What problem does it solve?
It turns a structured tasks.json plan into working code by resolving dependencies and delegating each task to the right expert agents, so you don’t have to coordinate complex parallel work manually.
Core Features & Use Cases
- Dependency-aware execution: Validates the tasks.json DAG and computes a topological execution order so prerequisite tasks complete before dependents start.
- Mixture of Experts delegation: For each ready task, selects and coordinates expert perspectives (from pre-assigned agent/moeExperts when available) to drive implementation and review.
- Acceptance-criteria verification: Ensures each task is marked complete only after the defined acceptance criteria pass, reducing “done-but-broken” outcomes.
Quick Start
Tell the AI: "Implement the tasks" (with a tasks.json present in the working directory) to validate the plan, resolve dependencies, and delegate each task to expert agents for implementation.